Information leadership in the advanced Asia–Paciﬁc stock

ific

turn,

c stockship of

oraneousAsians of

aryingnd ineffectsd

J. Japanese Int. Economies 19 (2005) 338–365

www.elsevier.com/locate/jjie

Information leadership in the advanced Asia–Pacstock markets: Return, volatility and volumeinformation spillovers from the US and Japan

Suk-Joong Kim

School of Banking and Finance, The University of New South Wales, Sydney, NSW 2052, Australia

Received 10 August 2003; revised 24 March 2004

Available online 6 May 2004

Kim, Suk-Joong—Information leadership in the advanced Asia–Pacific stock markets: Revolatility and volume information spillovers from the US and Japan

This paper investigates the nature of the stock market linkages in the advanced Asia–Pacifimarkets of Australia, Hong Kong, Japan and Singapore with the US and the information leaderthe US and Japan in the region since the early 1990s. It has been found that both the contempreturn and volatility linkages were significant and tended to be more intense after the 1997crisis period. However, the investigation of the dynamic information spillover effects in termreturns, volatility and trading volume from the US and Japan did not produce such time-vinfluence. In general, significant dynamic information spillover effects from the US were fouall the Asia–Pacific markets, but the Japanese information flows were relatively weak and thewere country specific.J. Japanese Int. Economies 19 (3) (2005) 338–365. School of Banking anFinance, The University of New South Wales, Sydney, NSW 2052, Australia. 2004 Elsevier Inc. All rights reserved.

JEL classification: G15; G14

Keywords: Information spillover; Asia–Pacific stock markets; Trade volume

E-mail address: [email protected].

0889-1583/$ – see front matter 2004 Elsevier Inc. All rights reserved.doi:10.1016/j.jjie.2004.03.002

http://www.elsevier.com/locate/jjie

mailto:[email protected]

S.-J. Kim / J. Japanese Int. Economies 19 (2005) 338–365 339

is wellelationsicingnshipslatilityee,

rkets.ding

ed af-,nd thele of

et beingan and

ence.e lib-ationsd987

ed by;, 1998;a–ries in

USt Asian000;latee stockAsian

many

trading

panese

1. Introduction

The existence of financial market linkages amongst advanced equity marketsdocumented. Numerous researchers find significant contemporaneous return corrwhich is not surprising considering the implications of international capital asset prmodels. In addition, dynamic market interdependences which indicate causal relatiowere also investigated by many researchers who report significant price and vospillovers between advanced markets (inter aliaHamao et al., 1990; Theodossiou and L1993; Koutmos and Booth, 1995; Connolly and Wang, 2000; Bae et al., 2000). A com-mon finding in these studies is the role of the US market in leading other major maIn addition to return and volatility spillovers, the information content of the US travolume had a significant causal influence in other markets (for example, seeLee andRui, 2002).1 These contemporaneous and dynamic inter-market linkages intensifiter the 1987 global stock market crash.Arshanapalli and Doukas (1993), among othersfind enhanced market linkages with increasing US influence on French, German aUK markets for the post-crisis period. However, the literature reports a negligible rothe Japanese market in information leadership, despite the Japanese stock markworld’s second largest, and an absence of significant market linkages between Japother major markets of the US and Western Europe.2

Asia–Pacific financial markets also exhibit significant and growing interdependThe increasing regionalisation of economic activities since the mid-1980s and theralisation of stock markets from late 1980s resulted in regional economic integr(Phylaktis, 1997, 1999)and growing stock market interdependence(Janakiramanan anLamba, 1998; Pan et al., 1999). Market linkages are noticeably greater for the post-1period as reported inArshanapalli et al. (1995), and for the post-1997 period(Chow, 1999;Kaminsky and Schmukler, 1999; Girard and Rahman, 2002). In addition, informationleadership of the US market is confirmed in the Asia–Pacific markets as evidencsignificant first and second moment return spillovers (inter aliaArshanapalli et al., 1995Liu et al., 1996; Liu and Pan, 1997; Ghosh et al., 1999; Janakiramanan and LambaGirard and Rahman, 2002). Another potential source of information flow for the AsiPacific markets is Japan due to its economic linkages with the rest of the countthe region. A number of studies report significant spillover effects from both theand the Japanese markets to the Asia–Pacific markets especially since the Easfinancial crisis of 1997(Liu and Pan, 1997; Cha and Cheung, 1998; Cha and Oh, 2Ng, 2000). However, despite close economic linkages (especially from the mid to1980s) between Japan and other regional countries, the influence of the Japanesmarket had not been very strong until the onslaught of the financial crises in the Eastcountries in 1997(Chow, 1999; Cha and Oh, 2000).

1 The relationships among return, volatility and trading volume within a market were investigated byresearchers. Significant contemporaneous linkages between return and trading volume are reported inTauchenand Pitt (1983) and Karpoff (1987), and between variance (mostly measured as absolute price change) andvolume byKarpoff (1987) and Gallant et al. (1992). In addition, dynamic causal relationships are reported inLeeand Rui (2002), Chen et al. (2001) and Chordia and Swaminathan (2000).

2 Bae and Karolyi (1994), however, report that the degree of market linkages between the US and the Ja
stock markets are significantly understated when good and bad market returns are not investigated separately.

340 S.-J. Kim / J. Japanese Int. Economies 19 (2005) 338–365

nd at, and. Dueaggre-

betterstiga-illover, the in-prove

e ex-

sion ofUSddressongst

he rolemajor

ancedositive,of

ntry-

efore

orane-

in the

tasented.s and

-radailyn-

ing the

The existing body of studies concentrates mostly on weekly return horizons (abest, daily) in their investigations of market linkages and information spillover effectsit rarely go beyond the examination of the first and second moment spillover effectsto the existence of trading time differences between the US and the Asia–Pacific, disgating the daily return horizons into overnight and intradaily periods would produceinsights into the nature of the market interdependence as this would allow the invetion of contemporaneous (co-movements) and dynamic (causation) information speffects. This research angle has been neglected by many researchers. In additionformation content of the US and Japanese market trading volumes would potentiallyuseful in providing additional tradable information for the Asia–Pacific markets. To thtent that volume information can be used to infer future stock returns(Blume et al., 1994)and that US and Japanese market returns lead other Asia–Pacific markets, the incluthe volume information in the analysis would allow much richer investigation of theand Japanese stock market leadership in the region. The aim of this paper is to athese issues. Specifically, the time varying nature of the stock market linkages amadvanced stock markets in the Asia–Pacific region and the US is investigated and tof the information leadership of the US and Japan in the region is examined. Thefindings of the paper are:

(i) contemporaneous return and volatility correlations amongst the US and the advAsia–Pacific stock markets of Australia, Hong Kong, Japan and Singapore are pand significant, and are considerably higher in the post-1997 Asian crisis period

(ii) in addition to return and volatility spillovers, significant dynamic spillover effectsthe US trading volume are found in all countries,

(iii) dynamic information spillover effects from Japan are generally weak and couspecific, and

(iv) there is no evidence of significant difference in the dynamic spillover effects band after the 1997 Asian crisis period.

Thus, this paper makes following contributions to the existing literature:

(i) provision of updated and comprehensive evidence on the nature of the contempous and dynamic stock market linkages in the region, and

(ii) new evidence that sheds light on the information leadership of the US and Japanregion.

The rest of this paper is organised as follows.Section 2provides the details of the daused in the paper and the results of the preliminary analysis of the data are preSection 3contains the analyses of contemporaneous correlations of the daily returnvolatilities amongst the five countries under investigation.Section 4provides a further evidence of market linkages in terms of causal influences of the US and Japanese intreturns, volatilities and trading volumes.Section 5presents the investigations of the cotemporaneous and dynamic information spillover effects from the US and Japan us
EGARCH modelling methodologies. Finally, conclusions are presented inSection 5.


k mar-, high,anese

riod 24aitsr-tweenarketsded asove-

whends forf con-ily (D),dayonay

sy

arketsdnce isistentds ofwnesswhichurtosisiod inations,

-linearion inencengle–re thusseriale re-ssed in

TOPIX

2. Data and preliminary analysis

The stock markets investigated are the US and four advanced Asia–Pacific stockets of Australia, Japan, Hong Kong and Singapore. Daily index observations (openlow and close) for the five markets and the trading volumes of the US and the Japmarkets were obtained from Commodities Systems, Inc. and Datastream for the peJuly 1990 to 27 March 2002.3 The indexes are All Ordinaries, TOPIX, Hang Seng, StrTimes and S&P 500, respectively for each country.Figure 1shows the time line of the maket trading hours of the Asia–Pacific and the US markets. While there are overlaps betrading hours of the Asia–Pacific markets, the US market is closed when the other mare operating. The information flow from the Japanese market, which can be regara regional information, is thus contemporaneous while the (overnight) US market mments, which constitute global information, can influence the Asia–Pacific marketsthey open three to four hours after the close of the US market. Various holding perioreturns and volatilities can be constructed with a view to ascertaining the nature otemporaneous and lead-lag relationships amongst the five markets. These are daovernight (ON) and intradaily (ID). The daily return period is from closing price ont − 1 to closing price on dayt , which envelopes the overnight return period (closingday t − 1 to opening on dayt) and the intradaily return period (opening to closing on dt). The daily and overnight return periods on (calendar) dayt in the Asia–Pacific marketoverlap with the US intradaily return period on dayt − 1, whereas intradaily return on dat periods do not (seeFig. 1).

The summary statistics of the market returns of the US and the Asia–Pacific mover the three holding periods are shown inTable 1. The mean is fairly close to zero anthe daily variance is spread over the two sub-holding periods in all cases. The variahigher during the intradaily period than the overnight period in all cases which is conswith the established empirical observation that points to higher volatility during periomarket trading than periods over which market was not trading. The negative skeis observed in all cases, except for the intradaily period for Japan and Singapore,is the usual finding for the financial return series. Both the skewness and excess kare significantly larger in magnitude over the overnight period than the intradaily perall cases except for Australia. This suggests a higher frequency of extreme observespecially negative ones, during the overnight return period. Both linear and nonserial correlations are highly significant in all cases except for the non-linear correlatAustralia and the US for the overnight period. In addition, there is a significant evidof asymmetric responses of the volatility to innovations as shown by the significant ENg statistics in all cases. The common characteristics of the index returns overall asignificant negative skewness, leptokurtosis, highly significant linear and non-linearcorrelation, and asymmetric responses of volatility to innovations. Modelling of thesturns must account for these characteristics. The modelling issues are further discuSection 5.

3 The choice of the starting point of the sample was due to the unavailability of trade volume data for the
prior to this date.

342S.-J.K

im/J.Japanese

Int.Econom

ies19

(2005)338–365

att ; ID: Intradaily: Opening att to Closing att ; D: Daily:US: US stock market.

Fig. 1. Stock market trading hours in the Asia–Pacific and the US. ON: Overnight: Closing att − 1 to OpeningClosing att − 1 to Closing att ; AP: Asia–Pacific markets of Australia, Hong Kong, Japan and Singapore;

S.-J.Kim

/J.JapaneseInt.E

conomies

19(2005)

338–365343

pore US

Overnight(b) Intradaily(c) Daily(a) Overnight(b) Intradaily(c)

9 0.0274 −0.0064 0.0430 −0.0039 0.04662 0.6298 1.1393 1.1011 0.0300 1.12045 −2.0322 0.3693 −2.6405 −2.8198 −2.5700

45.74 14.07 55.99 127.15 54.00

** 62.61*** 114.80*** 49.27*** 52.88*** 53.54***

0} {0.0000} {0.0000} {0.0003} {0.0001} {0.0001}** 103.75*** 1667.00*** 644.83*** 25.94 721.01***

0} {0.0000} {0.0000} {0.0000} {0.1678} {0.0000}** 110.14*** 322.74*** 111.21*** 36.74*** 141.29***

0} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000}the brackets arep-values.

on for squared returns; E–N:Engle and Ng’s (1993)joint

Table 1Statistical properties of overnight and daily stock market index return

Australia Japan Hong Kong Singa

Daily(a) Overnight(b) Intradaily(c) Daily(a) Overnight(b) Intradaily(c) Daily(a) Overnight(b) Intradaily(c) Daily(a)

Summary statisticsMean 0.0340 −0.0016 0.0357 0.0037 0.0032 −0.0285 0.0366 0.0194 0.0172 0.020Variance 0.9611 0.0681 0.8378 1.4511 0.1381 1.3931 3.3847 1.3651 1.7324 1.744Skewness−6.4499 −2.6700 −7.1249 −0.2851 −0.3054 0.2937 −3.4459 −6.0933 −0.6803 −0.437Kurtosis 172.15 49.75 203.70 11.34 15.07 4.38 73.82 157.18 8.05 15.06

Tests of iid(d)(d)Q(20) 47.25*** 83.58*** 43.91*** 68.12*** 53.72*** 92.11*** 47.42*** 56.57*** 54.39*** 102.89*

{0.0005} {0.0000} {0.0015} {0.0000} {0.0001} {0.0000} {0.0005} {0.0000} {0.0001} {0.000Q2(20) 147.37*** 17.88 473.52*** 406.95*** 195.07*** 441.05*** 958.18*** 353.65*** 1178.40*** 444.59*

{0.0000} {0.5954} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.000E–N 104.62*** 19.46*** 213.26*** 67.65*** 162.62*** 109.72*** 316.97*** 264.03*** 177.15*** 119.96*

{0.0000} {0.0002} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.000

Stock market indexes are All Ordinaries, TOPIX, Hang Seng, Straits Times, and S&P 500. Numbers in(a) ln(P Close

t /P Closet−1 ) × 100.

(b) ln(POpent /P Close

t−1 ) × 100.

(c) ln(P Closet /P

Opent−1 ) × 100.

(d) Q(20): Box–PierceQ-test of serial correlation for returns;Q2(20): Box–PierceQ-test of serial correlatisign bias test.*** Significance at the 1% level.


kagesstruc-ee andf suchntem-for all

i.e.

d mo-sitivethem.

rangere. Thetweenapore,ralia–turalsub-t-crisisre of

is pe-amples.in the

ationore. Thea 58%latilitycloselyfourthinst thewhich

rrela-of thed thatPacific

terest-sality,

inves-

3. Correlation analysis

The most common and also the simplest method of investigating market linadopted by early researchers is to examine the return and volatility correlationtures amongst the markets under investigation. The aim is to ascertain the degrthe nature of market co-movements without necessarily investigating the drivers oco-movements. Ideally, this requires the holding periods for comparison to be coporaneous, and this can be accomplished by considering daily holding periodsmarkets (daily period on dayt for the Asia–Pacific and on dayt − 1 for the US mar-kets) which result in overlapping holding periods. The daily returns on dayt are calcu-lated as ln(P Close

t /P Closet−1 ) × 100, and the volatilities are simply the squared returns,

[ln(P Closet /P Close

t−1 ) × 100]2.Table 2reports the contemporaneous correlation analysis of the first and secon

ments of daily stock market returns. Both the return and volatility correlations are poin all cases indicating the presence of some common global information that droveFor the full sample (shown in the first panel), the bilateral correlations of returnsfrom 0.3016 between the US and Japan to 0.5531 between Hong Kong and Singapovolatility correlations range from 0.1882 between the US and Japan to 0.6122 beAustralia and Singapore. The top three pairs in terms of size are Hong Kong–SingAustralia–US and Australia–Hong Kong for returns, and Australia–Singapore, AustHong Kong and Hong Kong–Singapore for volatilities. In order to account for the strucbreak around the 1997 currency crisis, the full sample (full-sample) is divided into twosamples, the pre-crisis period of 24 July 1990 to 30 June 1997 (sample 1) and the posperiod of 1 July 1997 to 27 March 2002 (sample 2), and then the time-varying natumarket linkages in the region is investigated. The second and the third panels ofTable 2show the return and volatility correlation linkages during the pre- and post-1997 crisriods, respectively. The last panel reports percentage changes over the two subsComparing samples 1 and 2, there is a noticeable rise in the bilateral correlationspost-crisis period in all cases except for a sizable drop of 29% in the volatility correlbetween Japan and Singapore and a marginal 1% drop between the US and Singaplargest increase in return correlation is shown between Hong Kong and Japan withrise, closely followed by a 48% rise between Australia and Japan. The highest vocorrelation rise is between Australia and Singapore with as much as a 112% rise,followed by a 108% increase between the US and Hong Kong. The last row of thepanel shows the average changes in the bilateral correlations for each market agaother four. The highest average rise in return correlation is against Japan (41% rise)shows the rising importance of Japan in the region. The highest for the volatility cotion, however, is against Hong Kong (58% rise) which might suggest the sensitivityHong Kong’s market to the volatility of the other markets. In short, it has been revealeboth the first and second moment linkages in market returns in the advanced Asia–markets and the US market were significant and grew over time.

Although the correlation analysis of contemporaneous market returns revealed ining results regarding the extent of market linkages in the region, the question of cauthat is the direction of information flows, can not be addressed. We turn now to the
tigations of the causal flows of information in the region.

S.-J.Kim

/J.JapaneseInt.E

conomies

19(2005)

338–365345

y correlations

Japan Hong Kong Singapore US

0.2134 0.6043 0.6122 0.36701 0.1902 0.1892 0.18820.1902 1 0.5506 0.39790.1892 0.5506 1 0.22880.1882 0.3979 0.2288 1

0.2123 0.3958 0.3083 0.30781 0.1861 0.2727 0.17160.1861 1 0.3978 0.19580.2727 0.3978 1 0.20610.1716 0.1958 0.2061 1

0.2627 0.6356 0.6531 0.37431 0.2262 0.1923 0.22670.2262 1 0.5607 0.40760.1923 0.5607 1 0.20320.2267 0.4076 0.2032 1

-crisis correlations24 61 112 22

0 22 −29 3222 0 41 108

29 41 0 −132 108 −1 0

12 58 30 40

Table 2Correlations of market return and volatility

Daily returns correlations Daily volatilit

Australia Japan Hong Kong Singapore US Australia

Full-sample: 24 July 1990 to 27 March 2002Australia 1 0.3732 0.4680 0.4027 0.4838 1Japan 0.3732 1 0.3406 0.3255 0.3016 0.2134Hong Kong 0.4680 0.3406 1 0.5531 0.3800 0.6043Singapore 0.4027 0.3255 0.5531 1 0.3334 0.6122US 0.4838 0.3016 0.3800 0.3334 1 0.3670

Pre-crisis: 24 July 1990 to 30 June 1997Australia 1 0.3036 0.3904 0.3282 0.4482 1Japan 0.3036 1 0.2609 0.3076 0.2458 0.2123Hong Kong 0.3904 0.2609 1 0.4450 0.3333 0.3958Singapore 0.3282 0.3076 0.4450 1 0.2931 0.3083US 0.4482 0.2458 0.3333 0.2931 1 0.3078

Post-crisis: 1 July 1997 to 27 March 2002Australia 1 0.4493 0.5395 0.4705 0.5255 1Japan 0.4493 1 0.4126 0.3522 0.3534 0.2627Hong Kong 0.5395 0.4126 1 0.6097 0.4049 0.6356Singapore 0.4705 0.3522 0.6097 1 0.3525 0.6531US 0.5255 0.3534 0.4049 0.3525 1 0.3743

Percentage difference between post-crisis correlations and preAustralia 0 48 38 43 17 0Japan 48 0 58 14 44 24Hong Kong 38 58 0 37 21 61Singapore 43 14 37 0 20 112 −US 17 44 21 20 0 22

Average change 37 41 39 29 26 54

Daily returns: ln(P Closet /P Close

t−1 ) × 100; daily volatility:[ln(P Closet /P Close

t−1 ) × 100]2.


re duetigationationusality

behinder

-

(

-ausings un-radailyted ass in

los-ones

-tility,

p withce the

l influ-ificant

in

4. Direction of information flows

It is not clear whether the significant first and second moment return correlations ato contemporaneous factors or due to dynamic causal relationships. Thus, an invesof causal flow of information in the region is required to ascertain the nature of informleadership of the US and Japan in the Asia–Pacific region. In this section, Granger catests are performed to ascertain the causal direction of information flows. The ideathis is the premise that if eventx occurs before eventy and forecasts ofy are more accuratwith the past values ofx in the information set than without, thenx is said to Grangecausey. Formerly, a test equation is written as

(1)yt = a +m∑

i=1

bi · yt−i +n∑

j=1

cj · xt−i + et

and the Granger causality fromx to y, written asxG.C.−→ y, is tested as a test of joint sig

nificance ofcj s, i.e. H0: x does not Granger causey is tested as H0: cj = 0 for all j . Inthe current analysis, the US and Japanese market returns and trading volumes (x variables)are tested to see if they Granger cause the market returns of the other marketsy vari-ables), and also the US and Japanese market volatilities and trading volumes (x variables)are tested for their effect on the market volatilities of the other markets (y variables).The bi-directional causation is tested only between the US–Japan pair,4 and only the onedirectional test is carried out for the smaller markets (the US and Japan Granger cAustralia, Hong Kong and Singapore). The holding periods for returns and volatilitieder consideration are chosen to eliminate overlaps and this is done by considering int(ID) holding periods with appropriate lags in all cases. The returns are then calculaRID,t = ln(P Close

t /POpent ) × 100. Instead of using the squared returns for volatilities a

the previous section, the intradaily volatility is calculated as:

VT ID,t = 1

2· [ln(

PHight

) − ln(P Low

t

)]2

(2)− (2 ln(2) − 1

) · [ln(P

Opent

) − ln(P Close

t

)]2.

This Garman–Klass volatility utilizes all four daily price observations of opening, cing, high and low, and is shown to have a significant efficiency gain over a simplersuch as squared or absolute price changes(Garman and Klass, 1980). It also has the advantage, compared to alternative measures of daily volatility such as GARCH volaof being pre-determined and observable by market participants at timet with a day’slag. The trading volume has been shown to have a positive feedback relationshireturn volatilities in the US and Japan, and the US volume has shown to influenJapanese market variables (Lee and Rui, 2002; see alsofootnote 1). Thus, it is of greatinterest to investigate whether the US and Japanese volume have similar causaence on the other Asia–Pacific markets. Both trading volume series contain sign

4 Evidence of bi-directional linkages in terms of return spillovers between the two markets are reportedBae
and Karolyi (1994), Karolyi and Stulz (1996), and Connolly and Wang (2000).


uation

ngthyl mar-inally

es (atsalitye in-

andthedditionf mar-ection,er in

ation

eral,nificantUS andtility.

ginallye USort, ther Asia–

apanese

vol-istently

he pres-S, and

es the

th, interest

re is no

similarest that,uld the

positive (linear and non-linear) trends, and so the residuals from the detrending eqVolumet = a + b · T + c · T 2 + et are used as detrended volumes, denoted asVMID,t .5

The left half ofTable 3presents the US Granger causality test results. The lag lechosen is 5 days (m = n = 5) which represent one week.6 For the full sample, the intradailUS return and volatility Granger caused the returns and volatilities, respectively, in alkets. In addition, the US volume Granger caused the returns in Australia (only margat 10%) and Hong Kong (significant at 5%), and Granger caused intradaily volatilitileast at 5%) of all the Asia–Pacific markets. While a similar pattern of Granger cauis revealed for the US return and volatility in both sample 1 and sample 2, the volumformation was more relevant in sample 1.7 The US volume Granger caused the returnsvolatilities of Australia, Japan and Hong Kong in sample 1, while only the volatility inSingapore was Granger caused in sample 2. Thus, the evidence indicates that in ato the significant contemporaneous correlations in the first and second moments oket returns between the US and the Asia–Pacific markets reported in the previous sthere is a dynamic causal relationship between them confirming the role of the formgenerating global information that affects the latter. That is, the US market informleadership is confirmed in the Asia–Pacific.

The right half ofTable 3reports the tests of Granger causality from Japan. In genthe Japanese market information failed to Granger cause other markets to any sigextent. For the full sample, the Japanese return Granger caused the returns in theHong Kong, while only the US volatility was Granger caused by the Japanese volaThe Japanese volume data had very little influence in all cases, except for the marsignificant (at 10%) influence on the returns in Hong Kong and Singapore. As in thinfluences above, sample 1 has more significant causal influence, in general. In shJapanese market movements failed to have a significant causal influence in the othePacific markets, however, the US market was shown to be Granger caused by the Jinformation.

In sum, the evidence shows that the US market (returns, volatility and tradingume) Granger caused Asia–Pacific markets, whereas only the US market was consGranger caused by the Japanese market information, in general.

5. Information spillovers

The correlation analysis and the Granger causality tests reported above showed tence of contemporaneous market linkages in the advanced Asia–Pacific and the Usignificant Granger causalities from the US market. This section further investigat

5 Although the volume series do not contain unit roots, significant trends are detected in both cases. Bob andc are statistically significant. To save space, details of the detrending analysis are not reported, howeverreaders can obtained them from the author upon request.

6 In addition, BIC was used to determine the optimal lag length for each testing equation. However, thenoticeable difference in the results compared to the ones reported here.

7 This may be due to the increasing importance of common shocks that affect all the markets in theway as evidenced in the previous section by the increasing correlations in sample 2. This might suggin relative terms, idiosyncratic shocks emanating from the US are reduced in importance, and so wo
information contained in the US market volume movements.

348S.-J.K

im/J.Japanese

Int.Econom

ies19

(2005)338–365

Japan Granger causes

Australia HK SP US

5.57 10.88* 3.20 26.75***

0.3501} {0.0539} {0.6687} {0.0001}4.32 7.40 6.84 11.98**

0.5039} {0.1926} {0.2329} {0.0350}3.67 9.47* 9.83* 8.440.5978} {0.0918} {0.0802} {0.1334}5.24 3.42 7.37 8.560.3868} {0.6355} {0.1946} {0.1281}

3.76 14.73** 5.92 15.94***

0.5837} {0.0116} {0.3142} {0.0070}5.40 3.78 10.52* 10.27*

0.3694} {0.5808} {0.0618} {0.0678}3.42 5.48 4.09 4.520.6351} {0.3604} {0.5370} {0.4772}5.21 2.42 4.91 5.200.3904} {0.7889} {0.4270} {0.3919}

(continued on next page)

Table 3Tests of Granger causality of the US and Japanese stock market information

yt = a +m∑

i=1

bi · yt−i +n∑

j=1

cj · xt−i + et , H0: xdoes not Granger causey, cj = 0 for all j

k = US Granger causes k =x

G.C.−→ y Australia Japan HK SP xG.C.−→ y

Full-sample: 24 July 1990 to 27 March 2002

RUSID

G.C.−→ RkID 158.10*** 25.20*** 15.31*** 15.54*** RJP

IDG.C.−→ Rk

ID{0.0000} {0.0000} {0.0000} {0.0000} {VTUS

IDG.C.−→ VTk

ID 27.49*** 20.57*** 5.12*** 10.83*** VTJPID

G.C.−→ VTkID{0.0000} {0.0000} {0.0001} {0.0000} {

VMUSID

G.C.−→ RkID 1.87* 1.62 2.25** 0.88 VMJP

IDG.C.−→ Rk

ID{0.0960} {0.1510} {0.0467} {0.4927} {VMUS

IDG.C.−→ VTk

ID 2.37** 4.34*** 3.31*** 3.39*** VMJPID

G.C.−→ VTkID{0.0369} {0.0006} {0.0056} {0.0047} {

Pre-crisis: 24 July 1990 to 30 June 1997

RUSID

G.C.−→ RkID 58.02*** 14.36*** 18.73*** 9.48*** RJP

IDG.C.−→ Rk

ID{0.0000} {0.0000} {0.0000} {0.0000} {VTUS

IDG.C.−→ VTk

ID 10.52*** 5.30*** 12.45*** 1.09 VTJPID

G.C.−→ VTkID{0.0000} {0.0001} {0.0000} {0.3645} {

VMUSID

G.C.−→ RkID 2.77** 3.65*** 1.86* 1.09 VMJP

IDG.C.−→ Rk

ID{0.0168} {0.0027} {0.0987} {0.3639} {VMUS

IDG.C.−→ VTk

ID 4.97*** 4.15*** 4.52*** 1.10 VMJPID

G.C.−→ VTkID{0.0002} {0.0009} {0.0004} {0.3591} {

S.-J.Kim

/J.JapaneseInt.E

conomies

19(2005)

338–365349

Japan Granger causes

Australia HK SP US

5.32 7.26 5.05 13.96**

0.3783} {0.2023} {0.4097} {0.0159}6.19 6.92 6.50 7.130.2879} {0.2270} {0.2605} {0.2110}8.09 5.51 8.93 11.34**

0.1513} {0.3565} {0.1117} {0.0451}4.74 2.47 5.03 5.680.4486} {0.7813} {0.4117} {0.3388}Japan, Hong Kong and Singapore} for the US causality,olding period of open to close att , m = n = 5. Test statistics are

Table 3 (Continued)

k = US Granger causes k =x

G.C.−→ y Australia Japan HK SP xG.C.−→ y

Post-crisis: 1 July 1997 to 27 March 2002

RUSID

G.C.−→ RkID 96.18*** 12.95*** 3.97*** 6.92*** RJP

IDG.C.−→ Rk

ID{0.0000} {0.0000} {0.0014} {0.0000} {VTUS

IDG.C.−→ VTk

ID 10.52*** 14.79*** 1.22 3.47*** VTJPID

G.C.−→ VTkID{0.0000} {0.0000} {0.2982} {0.0040} {

VMUSID

G.C.−→ RkID 0.39 0.52 0.88 0.39 VMJP

IDG.C.−→ Rk

ID{0.8533} {0.7590} {0.4927} {0.8536} {VMUS

IDG.C.−→ VTk

ID 1.54 1.75 0.99 4.47*** VMJPID

G.C.−→ VTkID{0.1735} {0.1203} {0.4217} {0.0005} {

R – returns,VT – volatility, VM – detrended trading volume. Superscripts: US – US market,k = {Australia,and {Australia, Hong Kong, Singapore and US} for the Japanese causality. Subscript: ID – intradaily hreported with associatedp-values in brackets.

* Significance at the 10% level.** Idem., 5%.*** Idem., 1%.


cy fi-tions inith an

ll (seearsi-series

ation,obser-

iately

turns,above

ets are

returniod oniodseem-s

ation

linkages and the nature of information leadership in the region. As reported inTable 1,returns over various holding periods exhibit characteristics common to high frequennancial series: significant skewness, excess kurtosis and significant serial correlathe second moments. Various researchers find that exponential GARCH models wappropriate distributional assumption explain the daily stock price movements weBollerslev et al., 1992for an extensive survey of empirical papers). In this section, pmonious MA (moving average)–EGARCH(1,1) models are used to model the returnwith asymmetric response characteristics and they are shown as:

(3a)

yt = αc + αHOL · HOLt + εt +q∑

k=1

αk · εt−k,

εt = zt

√ht ∼ (0, ht ), zt ∼ iid(0,1),

(3b)

lnht = βc + βHOL · HOLt + βh · lnht−1 + βε1 · εt−1√ht−1

+ βε2 ·( |εt−1|√

ht−1−

√2

π

),

whereq is the number of lagged innovations required to remove residual serial correlHOLt is seasonal dummy that takes the number of days between two successivevations: 1 for normal weekdays, 3 for Mondays, and 2 or higher for days immedfollowing market closures due to holidays.

The US and Japanese market information with appropriate lags (intradaily revolatilities and detrended volumes) are then included as exogenous variables in themodel. In order to shed more light on the contemporaneous linkages reported inSection 3,contemporaneous as well as dynamic spillover effects from these two major markinvestigated.

5.1. Contemporaneous spillover effects

Contemporaneous information spillovers are investigated by using overlappinghorizons between the Asia–Pacific and the US markets. The intradaily holding perday t − 1 for the US (US− IDt−1) is contemporaneous with the overnight holding peron dayt for the Asia–Pacific markets (AP− ONt , seeFig. 1). In the case of the Japaneintradaily period on dayt , the other Asia–Pacific markets’ intradaily periods are contporaneous on dayt (AP− IDt ), whereas the overnight period on dayt is contemporaneoufor the US market (US− ONt ).

The EGARCH models appropriate for investigating contemporaneous US informspillovers areEqs. (4a) and (4b):

(4a)Rt = αR_US · RUSID,t−1 + αVM_US · VMUS

ID,t−1 + M(·),(4b)lnht = βVT_US · VTUS

ID,t−1 + βVM_US · VMUSID,t−1 + V (·),

where

Rt – overnight returns in the Asia–Pacific markets on dayt , RAPON,t ,


ay

the in-n

ay

ay

ignif-orma-e usedining

lyin all

ic aspectoremple 1)ple andhich

nosticsries. Thethe

hend their

lnht – conditional variance of overnight returns in the Asia–Pacific markets on dt ,lnhAP

ON,t ,

RUSID,t−1 – US intradaily returns on dayt − 1,

VTUSID,t−1 – Garman–Klass (intradaily) volatility in the US market on dayt − 1,

VMUSID,t−1 – detrended trading volume in the US market on dayt − 1,

M(·) – right-hand side of the conditional varianceEq. (3a), andV (·) – right-hand side of the conditional varianceEq. (3b).

The contemporaneous Japanese information spillovers are investigated usingtradaily periods on dayt for the Asia–Pacific markets (AP− IDt ) and the overnight returperiod on dayt for the US (US− ONt ) (seeFig. 1):

(5a)Rt = αR_JP· RJPID,t + αVM_JP· VMJP

ID,t + M(·),(5b)lnht = βVT_JP· VTJP

ID,t + βVM_JP· VMJPID,t + V (·),

where

Rt – intradaily returns in the Asia–Pacific markets on dayt , RAPID,t ,

and overnight return in the US on dayt , RUSON,t ,

lnht – conditional variance of intradaily returns in the Asia–Pacific markets on dt ,lnhAP

ID,t ,and conditional variance of overnight returns in the Asia–Pacific markets on dt ,lnhUS

ON,t ,

VTJPID,t – Garman–Klass (intradaily) volatility in the Japanese market on dayt ,

VMJPID,t – detrended trading volume in the Japanese market on dayt .

The information spillover effects are then investigated by examining the sign and sicance of the exogenous variables included in (4) and (5) above. Specifically, the inftion content of the intradaily returns and the trading volumes of the US and Japan arto explain the returns, and the intradaily volatilities and the trading volumes for explathe conditional volatilities of the other markets.

Table 4reports the US spillover effects.8 Examining the full sample, the US intradaireturn had a significant and positive influence on the Asia–Pacific market returns

8 Other aspects of the estimations point to the usual properties of the EGARCH models. The asymmetrof the model is shown in all cases, i.e. negativeβε1; however, it is significant only in one instance, Singapin sample 1. Return was higher on days following market closures of more than one day in Australia (saand Hong Kong (full-sample and sample 1), whereas a lower return was observed for Singapore (full-samsample 1). The significant holiday effect found on the volatility is generally positive (i.e. higher volatility) wis consistent with the accepted empirical evidence of higher volatility following market closures. The diagsuggest that the EGARCH models were successful in addressing the characteristics of the return seskewness and kurtosis of the standardized residuals,zt , of the models are substantially reduced compared toones reported inTable 1in all cases, and the white noise property ofzt is confirmed, in general, in all cases. Tremaining serial correlations in the mean in Australian returns and in Singapore were caused by outliers a
removal reduced the serial correlations.

352S.-J.K

im/J.Japanese

Int.Econom

ies19

(2005)338–365

VMUSID,t−1 + V (·)

Singapore

re-1997 Post-1997 Full-sample Pre-1997 Post-1997** −0.0418 0.0530*** 0.0313*** 0.0870**

} {0.3157} {0.0000} {0.0000} {0.0494}*** 0.0240 −0.0182*** −0.0156*** −0.0305} {0.3947} {0.0000} {0.0000} {0.3506}*** 0.4464*** 0.1237*** 0.0211*** 0.2379***

} {0.0000} {0.0000} {0.0001} {0.0000}0.1223* 0.0373*** −0.0345*** 0.0331

} {0.0901} {0.0000} {0.0000} {0.8018}−0.9960*** −0.5749*** 0.0895** −1.0320***

} {0.0000} {0.0006} {0.0453} {0.0000}−0.0522 −0.0511 −0.0408** −0.0264

} {0.3259} {0.1593} {0.0341} {0.7338}*** 0.4176*** 0.4462*** 0.0982*** 0.4641***

} {0.0000} {0.0000} {0.0000} {0.0006}*** 0.1739** 0.8277*** 0.9940*** 0.7197***

} {0.0437} {0.0000} {0.0000} {0.0000}0.3441*** 0.1905*** −0.0360 0.2930*

} {0.0000} {0.0097} {0.1882} {0.0519}2.0972*** 1.7433*** −0.5524*** 2.6559***

} {0.0010} {0.0000} {0.0032} {0.0000}0.6767** 0.0927 0.3271 0.3746*

} {0.0289} {0.6873} {0.3643} {0.0565}(continued on next page)

Table 4US information spillovers—contemporaneous

RAPON,t

= αR_US · RUSID,t−1 + αVM_US · VMUS

ID,t−1 + M(·), lnhAPON,t

= βVT_US · VTUSID,t−1 + βVM_US ·

Australia Japan Hong Kong

Full-sample Pre-1997 Post-1997 Full-sample Pre-1997 Post-1997 Full-sample P

αc 0.0041*** −0.0113*** −0.0021 0.0068 −0.0005 −0.0019 −0.0449** −0.0330{0.0036} {0.0000} {0.9814} {0.6220} {0.9587} {0.8706} {0.0242} {0.0139

αHol −0.0017 0.0028** 0.0001 0.0001 0.0053 0.0004 0.0283** 0.0212{0.3146} {0.0338} {0.9993} {0.9912} {0.3312} {0.9525} {0.0362} {0.0013

αR_US 0.0092*** 0.0403*** −0.0001 0.1040*** 0.0704*** 0.1437*** 0.3268*** 0.2709{0.0019} {0.0000} {0.7874} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000

αVM_US 0.0207* 0.0004 0.0043 −0.0075 0.0334 −0.1654** 0.0260 −0.0083{0.0625} {0.9260} {0.8481} {0.7426} {0.5023} {0.0143} {0.6249} {0.8538

βc −2.9113*** 0.2899 −4.8076 −0.3775 −0.6957 0.1739 −0.3372* −0.4157{0.0000} {0.3247} {0.3445} {0.1847} {0.1239} {0.2257} {0.0924} {0.2881

βε1 −0.0144 −0.0024 0.0989 0.0044 −0.0491 −0.0063 −0.0539 −0.0734{0.2823} {0.9727} {0.8971} {0.9048} {0.5219} {0.8564} {0.1881} {0.3168

βε2 0.7494*** 0.4446*** 0.4023*** 0.4165*** 0.4538*** 0.1284** 0.3815*** 0.4861{0.0000} {0.0011} {0.0000} {0.0000} {0.0000} {0.0458} {0.0000} {0.0000

βh 0.2229*** 0.9264*** −0.0706 0.7997*** 0.6887*** 0.9784*** 0.8165*** 0.6764{0.0000} {0.0000} {0.9497} {0.0000} {0.0000} {0.0000} {0.0000} {0.0001

βHol 0.3363*** −0.3081 0.8092 −0.0372 −0.0220 −0.1361 0.0473 −0.0972{0.0000} {0.2235} {0.8299} {0.7081} {0.8686} {0.1599} {0.4928} {0.2961

βVT_US 0.7364** 0.2941 −7.3244 0.5886 −0.7874 0.0194 1.9586** 2.3127{0.0303} {0.8784} {0.9022} {0.1283} {0.4607} {0.8896} {0.0128} {0.1226

βVM_US 0.1965 1.4821 0.0249 0.2791 0.5314 0.3359 0.1460 −0.1324{0.5061} {0.1797} {0.9944} {0.3400} {0.3719} {0.3066} {0.5808} {0.6071

S.-J.Kim

/J.JapaneseInt.E

conomies

19(2005)

338–365353

Singapore

re-1997 Post-1997 Full-sample Pre-1997 Post-1997

6 10 10 10−0.36 −1.66 −2.85 −0.41

6.73 31.61 40.45 11.0922.21 38.13*** 23.14 22.94

} {0.3292} {0.0085} {0.2818} {0.2920}25.59 2.01 13.45 19.97

} {0.1796} {1.0000} {0.8570} {0.4598}1.83 7.72* 10.14** 2.19

} {0.6092} {0.0523} {0.0174} {0.5348}tion; Sk-z and Kur-z are skewness and kurtosis of the standardized

le and Ng’s joint sign bias test ofzt . The numbers in the

Table 4 (Continued)



q 8 6 6 4 4 4 6 6Sk-z −2.0541 −1.49 0.10 −0.25 −1.14 0.19 −0.60 −0.59Kur-z 45.4380 22.3521 68.41 25.26 31.13 16.67 11.57 12.41Q(20)-z 62.28*** 59.75*** 71.09*** 21.22 19.39 14.47 23.88 26.42

{0.0000} {0.0000} {0.0000} {0.3845} {0.4966} {0.8058} {0.2476} {0.1523Q(20)-z2 20.04 19.72 17.80 11.32 4.92 10.81 12.25 10.09

{0.4555} {0.4754} {0.6005} {0.9374} {0.9998} {0.9509} {0.9071} {0.9665E–N 1.1081 0.58 3.73 1.71 1.75 1.56 1.03 0.16

{0.7751} {0.9015} {0.2925} {0.6353} {0.6257} {0.6675} {0.7929} {0.9838

q is the number of error terms in the conditional mean equations included to reduce residual serial correlaresidualszt ; Q(20)-z andQ(20)-z2 are Box–PierceQ-test of serial correlation ofzt andz2

t ; E–N is the Engbrackets arep-values.



of theof

es be-ollar.

inan andr evi-fter thet forypearsacificlatil-

ingly,ntentarketratede US

n fullppar-returns

pondositiveg vol-fic asandrns incases of

mationtneousin sam-e theolatil-wasple).g up

ayle and

t drop int period

cases. The Hong Kong market was the most responsive judging by the magnitudemean spillover coefficientαR_US, 0.3268, which is at least up to three times the sizethose of the other markets. This is not surprising due to the close financial linkagtween Hong Kong and the US stemming from the currency peg of the Hong Kong DAs a confirmation of higher return correlations in the post-1997 sample reportedTa-ble 2, the contemporaneous mean spillover coefficient in sample 2 is doubled in JapHong Kong, and nearly ten fold increase is observed in Singapore. This is a cleadence of increasing market linkages between the US and the Asia–Pacific markets a1997 crisis. The volatility spillover is significant in all cases for the full sample excepJapan, and the positive coefficient,βVT_US, indicates a higher (lower) intradaily volatilitin the US had a significant market exciting (calming) effect in these markets. It apthat the trading environment that prevailed in the US was transmitted to the Asia–Pmarkets resulting in similar market movements. Similar to the return spillovers, the voity spillover was at its greatest in Hong Kong, closely followed by Singapore. Interesta significant negative spillover is shown in sample 1 in Singapore. The information coof the contemporaneous US market volatility apparently was helpful in resolving muncertainties leading to a lower return volatility in Singapore. However, this evapoin sample 2 and the positive relationship was dominant. Information contained in thtrading volume was useful in explaining the returns for Australia and Singapore isample and Hong Kong in sample 2. An increasing volume in the US market was aently seen as an encouraging sign, on average, and helped to significantly raise theof these two markets. This contrasts with the negative coefficient,αVM_US, found in Japanfor sample 2 and Singapore for sample 1. Although the market volatility did not resto the volume in any of the markets considered for the whole sample, a significant pinfluence is detected in Hong Kong and Singapore in sample 2. Higher US tradinumes obviously had the similar influence in raising the volatilities in the Asia–Pacidid higher US volatility. In general, contemporaneous intradaily US return, volatilitytrading volume moved the first and second moments of the Asia–Pacific market retuthe same direction, and the post-1997 sample reports increased responses in thereturns linkages.

Table 5reports the estimation results of the Japanese contemporaneous inforspillovers.9 The contemporaneous return spillovers, measured asαR_JP, had a significanpositive influence in all the markets except for the US and, as in the contemporaUS market influence discussed above, market linkages were considerable higherple 2. The US market, however, moved in the opposite direction in full sample, whilspillovers are individually insignificant in samples 1 and 2. The Japanese intradaily vity, βVT_JP, had a market calming influence, in general. The conditional volatilitysignificantly reduced in Singapore (full-sample and sample 1) and in the US (full-samApparently, the information content of the Japanese volatility was useful for clearin

9 The coefficient for the asymmetric effect,βε1, is now significant in most of the estimations. The holideffect is generally insignificant in the mean, except for significant negative impact in Singapore (full-sampsample 1), and a positive volatility response is reported in all cases except for the US where a significanmarket volatility is shown in all samples. The diagnostics suggest, in general, that the intradaily (overnigh
for the US) period characteristics were well modelled by the EGARCH models.

S.-J.Kim

/J.JapaneseInt.E

conomies

19(2005)

338–365355

JPID,t + βVM_JP· VMJP

ID,t + V (·)VM_JP· VMJP

ID,t + V (·)

US


4 −0.0190 −0.0196*** 0.0003 0.00225} {0.7287} {0.0000} {0.4765} {0.5371}7** −0.0068 0.0001 −0.0002 −0.00079} {0.7928} {0.3736} {0.4510} {0.3950}9 0.2673*** −0.0054*** −0.0007 0.00227} {0.0000} {0.0000} {0.1050} {0.3827}4 0.0859 0.0112 −0.0023* −0.0016**

7} {0.2040} {0.7478} {0.0701} {0.0189}4 −0.1519* 2.3114*** 4.3041*** 4.0076***

5} {0.0780} {0.0000} {0.0000} {0.0000}1 −0.0652*** −0.0100*** −0.1795** −1.4100***

8} {0.0024} {0.0012} {0.0202} {0.0000}6*** 0.2394*** 0.7913*** 0.1264*** 2.6590***

4} {0.0000} {0.0000} {0.0012} {0.0000}5*** 0.9639*** 0.9053*** 0.9880*** 0.5499***

0} {0.0000} {0.0000} {0.0000} {0.0000}8 0.1110* −1.6632*** −3.0270*** −2.8396***

9} {0.0559} {0.0000} {0.0000} {0.0000}3*** 0.2237* −0.0707* 0.1996 −0.28046} {0.0654} {0.0765} {0.2474} {0.5683}8 0.2765* −1.0569*** −2.6937*** −3.4014***

4} {0.0780} {0.0041} {0.0000} {0.0000}(continued on next page)

Table 5Japanese information spillovers—contemporaneous

For Australia, HK and SP: RAPID,t = αR_JP· RJP

ID,t + αVM_JP· VMJPID,t + M(·), lnhAP

ID,t = βVT_JP· VT

For the US: RUSON,t

= αR_JP· RJPID,t + αVM_JP· VMJP

ID,t + M(·), lnhUSON,t

= βVT_JP· VTJPID,t + β

Australia Hong Kong Singapore


αc 0.0181 0.0287 0.0037 0.1047** 0.1675** −0.0788 0.0307** 0.019{0.4248} {0.3955} {0.9207} {0.0062} {0.0002} {0.3054} {0.0295} {0.249

αHol 0.0090 0.0008 0.0191 −0.0266 −0.0377 0.0132 −0.0167** −0.013{0.5587} {0.9748} {0.4628} {0.2853} {0.1336} {0.7564} {0.0435} {0.010

αR_JP 0.1842*** 0.1621*** 0.2217*** 0.2686*** 0.2211*** 0.3533*** 0.0061** 0.001{0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0213} {0.449

αVM_JP −0.0531 −0.0082 −0.0962** 0.0569 0.0647 0.0821 −0.0255** −0.001{0.2895} {0.8909} {0.0421} {0.2303} {0.3124} {0.3941} {0.0411} {0.909

βc −0.4523* −0.2974* −0.4499** −0.0961** −0.2248*** −0.0256 −0.0068 −0.035{0.0732} {0.0595} {0.0243} {0.0120} {0.0067} {0.6400} {0.8969} {0.932

βε1 −0.0626* −0.0293 −0.0944* −0.0330*** −0.0389 −0.0440** −0.0723*** −0.191{0.0844} {0.2132} {0.0532} {0.0082} {0.1002} {0.0171} {0.0000} {0.494

βε2 0.2329** 0.1441*** 0.2548*** 0.1810*** 0.2117*** 0.1623*** 0.3714*** 0.690{0.0134} {0.0065} {0.0033} {0.0000} {0.0000} {0.0000} {0.0000} {0.028

βh 0.7308*** 0.9075*** 0.6554*** 0.9689*** 0.9474*** 0.9773*** 0.9771*** 0.966{0.0084} {0.0000} {0.0017} {0.0000} {0.0000} {0.0000} {0.0000} {0.000

βHol 0.1462*** 0.1508** 0.1015* 0.0725*** 0.1562*** 0.0248 0.0330 0.083{0.0015} {0.0352} {0.0817} {0.0096} {0.0070} {0.5130} {0.3786} {0.791

βVT_JP 0.7567 0.1461 1.1960** 0.0782 0.1195 0.0835 −0.2624*** −0.628{0.2358} {0.4090} {0.0212} {0.1320} {0.2047} {0.3216} {0.0077} {0.003

βVM_JP −0.0036 0.0809 −0.0112 0.2523 0.3806* 0.1244 0.7300*** 1.382{0.9709} {0.5309} {0.9382} {0.1302} {0.0727} {0.4050} {0.0000} {0.193

356S.-J.K

im/J.Japanese

Int.Econom

ies19

(2005)338–365

US


10 6 6 60.41 −1.31 −5.65 3.251.83 117.96 111.40 86.06

* 16.54 12.36 38.45*** 24.010} {0.6826} {0.9033} {0.0078} {0.2420}

18.64 5.13 25.54 29.10*

0} {0.5456} {0.9997} {0.1817} {0.0857}2.09 8.24** 2.19 1.70

4} {0.5543} {0.0414} {0.5346} {0.6364}

Table 5 (Continued)



q 6 6 6 4 4 4 10 10Sk-z −0.1344 −0.19 −0.11 −0.25 −0.40 0.08 3.34 5.09Kur-z 0.6193 1.0001 0.24 2.27 2.76 0.53 66.79 105.31Q(20)-z 21.42 19.76 16.10 16.68 16.49 10.62 54.47*** 185.89**

{0.3728} {0.4730} {0.7107} {0.6734} {0.6860} {0.9556} {0.0000} {0.000Q(20)-z2 27.93 12.09 14.94 12.08 12.60 17.51 0.31 0.25

{0.1111} {0.9128} {0.7798} {0.9134} {0.8941} {0.6195} {1.0000} {1.000E–N 1.0510 1.01 4.16 2.21 2.38 1.11 0.38 1.93

{0.7889} {0.7997} {0.2446} {0.5302} {0.4974} {0.7744} {0.9448} {0.587



had a

estingaving

ume tongrop isvealedshowanese

ifferentolatil-rently.en theies.

. There con-an are

e

ay

ilyd

heterogeneity of market sentiments in these two markets. The volume informationnegative influence on the market returns. A significant coefficient,αVM_JP, is observed forAustralia (sample 2), Singapore (full-sample) and the US (samples 1 and 2) suggthat market participants, on average, viewed a rise in trading activities in Japan ha negative influence on their markets. On the other hand, the evidence on the volvolatility spillovers,βVM_JP, is mixed. Higher volume caused higher volatility in HoKong (sample 2) and in Singapore (full-sample and sample 2), but a significant dobserved in the US (full-sample, sample 1 and sample 2). The investigation thus rethe followings. First, except for the mean spillover effects, where post-1997 samplesignificantly higher degree of market linkage, there is no obvious pattern of the Japinformation flows in the two subsamples. Second, the US market responses were dfrom the other markets in some instances (return to return spillovers and volume to vity spillovers) suggesting that the Japanese market information was interpreted diffeThis might be due to the different nature of economic (i.e. trade) relationship betweUS and Japan compared to that between Japan and the other Asia–Pacific econom

5.2. Dynamic spillover effects

Dynamic spillover effects from the US and Japan are investigated in this sectionreturn horizons that avoid overlaps between the Asia–Pacific and the US markets astructed and the influence of the lagged market information from the US and Japthen examined. The intradaily US and Japanese information att − 1 lead the intradailymarket movements in the Asia–Pacific on dayt (AP− IDt ), while the intradaily Japanesinformation att leads the intradaily US market movements att (US− IDt ).

The dynamic US information spillovers are then investigated using theEqs. (6a)and (6b)below for all the Asia–Pacific markets:

(6a)Rt = αR_US · RUSID,t−1 + αVM_US · VMUS

ID,t−1 + M(·),(6b)lnht = βVT_US · VTUS

ID,t−1 + βVM_US · VMUSID,t−1 + V (·),

where



ID,t .

The Japanese information spillovers on dayt − 1 are investigated using the intradaperiods on dayt for the Asia–Pacific markets (AP− IDt ) and the intradaily return perioon dayt − 1 for the US (US− IDt−1):

(7a)Rt = αR_JP· RJPID,t−1 + αVM_JP· VMJP

ID,t−1 + M(·),(7b)lnht = βVT_JP· VTJP

ID,t−1 + βVM_JP· VMJPID,t−1 + V (·),

where


ay

ortedses,forma-rkets.his issam-

nificantm theared toentem-

analy-lthoughtilityHonga signn theere theoundng and

d theencesin

e effecte non-

latilitydequacy

esent in

o haveUS are

specifi-r


and intradaily returns in the US on dayt − 1, RUSID,t−1,


ID,t ,

and conditional variance of overnight returns in the US on dayt − 1, lnhUSID,t−1.

The dynamic US market information spillover effects in the Asia–Pacific are repin Table 6.10 A positive and significant coefficient for the mean spillover effect in all caexcept for Singapore in sample 1, suggests that the lagged US return contained intion that was continued to be useful throughout the trading day in the Asia–Pacific maHowever, there is no evidence of more intense information transfers in sample 2. Tin contrast to the higher contemporaneous correlation in returns and volatilities inple 2 documented in previous sections. This suggests that although there is no sigincrease in importance of the mechanism that carries the idiosyncratic shocks froUS markets, common shocks, especially generated in the Asia–Pacific region, appehave become more important after the 1997 crisis episodes.11 For Australia and Japan, thspillover coefficient is considerable larger in all three samples compared to the coporaneous information transfers reported inTable 4. The volatility to volatility spillover isgenerally positive and significant in Australia and Japan. As in the contemporaneoussis, the Singaporean market shows different responses over the two subsamples. Athe volume to return spillover effect is generally insignificant, the volume to volaspillover is positive and significant in all cases, except for Japan in sample 1 andKong in sample 2. Higher trading volumes in the US were, in general, interpreted asof increasingly uncertain market condition which apparently had a similar impact iAsia–Pacific. This contrasts with the results for the contemporaneous analysis whvolume effect is not present, in general. Interestingly, significantly higher spillover is fin sample 2 for Australian and Japan, whereas the reverse is the case for Hong KoSingapore.

The dynamic information spillovers from Japan are reported inTable 7.12 The return toreturn spillover is significant and positive in Hong Kong (full-sample and sample 2) anUS (full-sample, sample 1 and sample 2); however, both positive and negative influare found for the other markets. The volatility to volatility spillover is significant onlyAustralia where opposite responses in sample 1 and sample 2 are shown. The volumon the mean is mixed and there is no evidence of a common response to it from th

10 The negative asymmetric effect of the innovation on the conditional variance and the significant voraising effect of market closures are significantly present in all cases. The diagnostics suggest the model ain all cases except for Singapore (full-sample and sample 1) where significant serial correlation is still prthe standardized residuals.11 Also, the role of idiosyncratic shocks from Hong Kong and Singapore in the US market appeared timproved in sample 2. The estimation results of dynamic spillover effects from these two countries to thenot reported in this paper to conserve space. Interested readers may obtain these results upon request.12 Except for the Australian market (full-sample and sample 2), there is no strong evidence of model miscation. In the case of Australia, the significant linear and non-linear serial correlations inzt were due to a numbe
of influential outliers.

S.-J.Kim

/J.JapaneseInt.E

conomies

19(2005)

338–365359

MUSID,t−1 + V (·)

Singapore


1*** −0.0988 −0.0285*** 0.0000 −0.04445} {0.1949} {0.0035} {0.9981} {0.5148}0* 0.0173 0.0140** 0.0000 −0.00791} {0.7162} {0.0133} {0.9930} {0.8539}3*** 0.1735*** −0.0015*** 0.0000 0.03630} {0.0000} {0.0002} {0.9988} {0.1385}5 0.1448 0.0174*** 0.0000 0.01941} {0.5161} {0.0012} {0.9929} {0.8304}3* −0.0279 0.0604 0.1703** −0.2044***

7} {0.4902} {0.3595} {0.0207} {0.0068}0** −0.0438** −0.0666*** −0.2762*** −0.0642***

5} {0.0266} {0.0001} {0.0000} {0.0005}1*** 0.1440*** 0.3092*** 0.4057*** 0.1890***

0} {0.0015} {0.0000} {0.0000} {0.0000}1*** 0.9789*** 0.9727*** 0.9843*** 0.9668***

0} {0.0000} {0.0000} {0.0000} {0.0000}4** 0.0252 −0.0236 0.1851*** 0.1430***

3} {0.3618} {0.6049} {0.0000} {0.0057}6 0.0815 −0.1095 −7.4512*** 0.1914**

2} {0.3526} {0.3005} {0.0000} {0.0295}8** 0.2122 0.6923*** 1.9712*** 1.0919***

1} {0.3594} {0.0000} {0.0000} {0.0002}(continued on next page)

Table 6US information spillovers—dynamic

RAPID,t = αR_US · RUS

ID,t−1 + αVM_US · VMUSID,t−1 + M(·), lnhAP

ID,t = βVT_US · VTUSID,t−1 + βVM_US · V



αc −0.0168 −0.0166 −0.0143 0.0034 0.0121 0.0060 0.0926** 0.151{0.4342} {0.5055} {0.7229} {0.9003} {0.7531} {0.9199} {0.0210} {0.000

αHol 0.0168 0.0156 0.0172 −0.0463** −0.0554*** −0.0293 −0.0364 −0.049{0.1820} {0.3129} {0.5478} {0.0114} {0.0088} {0.4365} {0.1181} {0.062

αR_US 0.3255*** 0.3442*** 0.3148*** 0.1923*** 0.2136*** 0.1799*** 0.2370*** 0.303{0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.000

αVM_US 0.0252 0.0859 −0.0652 0.0730 0.0624 −0.2165* 0.0442 −0.008{0.5517} {0.1073} {0.3934} {0.4800} {0.6074} {0.0742} {0.5689} {0.917

βc −0.5341*** −0.5086*** −0.1015 −0.1811*** −0.1387** −0.3349*** −0.0956** −0.162{0.0007} {0.0011} {0.4763} {0.0008} {0.0390} {0.0003} {0.0469} {0.050

βε1 −0.0220 0.0068 −0.1012*** −0.0876*** −0.1053*** −0.0357 −0.0450** −0.059{0.4549} {0.8057} {0.0005} {0.0000} {0.0001} {0.1872} {0.0200} {0.036

βε2 0.2468*** 0.2301*** 0.0441 0.2344*** 0.2397*** 0.1786*** 0.1772*** 0.190{0.0000} {0.0000} {0.3215} {0.0000} {0.0001} {0.0000} {0.0000} {0.000

βh 0.6747*** 0.6933*** 0.9686*** 0.9496*** 0.9660*** 0.8654*** 0.9637*** 0.951{0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.0000} {0.000

βHol 0.1414*** 0.1183** 0.0490 0.1224*** 0.0999** 0.2207*** 0.0721** 0.123{0.0041} {0.0196} {0.5758} {0.0005} {0.0234} {0.0004} {0.0410} {0.037

βVT_US 0.9877*** 1.8398*** 0.0568 0.2009** −0.0160 0.4231** 0.1388 −0.226{0.0097} {0.0066} {0.4858} {0.0467} {0.9470} {0.0139} {0.1786} {0.214

βVM_US 0.4808** 0.4365* 1.2551*** 0.3106** 0.2188 0.8312*** 0.3893*** 0.654{0.0223} {0.0647} {0.0001} {0.0136} {0.1973} {0.0002} {0.0012} {0.015

360S.-J.K

im/J.Japanese

Int.Econom

ies19

(2005)338–365

Singapore


12 10 10 100.02 0.32 0.20 0.320.53 3.45 93.04 1.824.16 43.81*** 29.95* 14.61

7} {0.9999} {0.0016} {0.0707} {0.7981}17.98 62.65*** 71.44*** 24.94

0} {0.5885} {0.0000} {0.0000} {0.2037}1.54 10.42** 1.23 1.58

4} {0.6732} {0.0153} {0.7456} {0.6637}

Table 6 (Continued)



q 8 4 600 4 4 4 12 12Sk-z −0.1052 −0.07 −0.19 0.17 0.32 0.01 −0.35 −0.46Kur-z 0.9022 0.6641 1.17 1.77 2.44 0.80 3.15 3.79Q(20)-z 5.58 15.58 9.23 19.64 8.88 25.61 9.28 12.07

{0.7424} {0.7424} {0.9801} {0.4808} {0.9842} {0.1791} {0.9794} {0.913Q(20)-z2 18.53 18.53 20.21 18.33 17.34 21.94 6.31 7.44

{0.5522} {0.5522} {0.4451} {0.5655} {0.6311} {0.3438} {0.9984} {0.995E–N 2.3034 2.30 3.73 1.58 2.62 0.39 0.83 1.51

{0.5119} {0.5119} {0.2923} {0.6647} {0.4541} {0.9414} {0.8423} {0.679


S.-J.Kim

/J.JapaneseInt.E

conomies

19(2005)

338–365361

· VTJPID,t−1 + βVM_JP· VMJP

ID,t−1 + V (·)_JP· VMJP

ID,t + V (·)

US

re-1997 Post-1997 Full-sample Pre-1997 Post-1997* 0.1190*** 0.0000 0.0140 −0.0042} {0.0005} {0.9989} {0.6515} {0.9482}

−0.0286** 0.0195 0.0203 0.0002} {0.0223} {0.2750} {0.3297} {0.9955}*** 0.0373*** 0.0663*** 0.0608*** 0.1230***

} {0.0000} {0.0000} {0.0001} {0.0000}*** 0.0449 −0.0629 −0.0973* 0.1247***

} {0.2449} {0.1400} {0.0639} {0.0012}−0.5486*** −0.0525 −0.0199 −0.0733

} {0.0039} {0.4244} {0.8385} {0.4128}* −0.1759*** −0.0853*** −0.0441** −0.2135***

} {0.0061} {0.0000} {0.0194} {0.0000}*** 0.6727*** 0.1237*** 0.0927*** 0.0902***

} {0.0000} {0.0000} {0.0003} {0.0069}*** 0.7183*** 0.9837*** 0.9862*** 0.9190***

} {0.0000} {0.0000} {0.0000} {0.0000}0.2371* 0.0372 0.0093 0.0625

} {0.0702} {0.4126} {0.8896} {0.3362}0.6929 −0.0120 −0.0037 0.1541

} {0.1932} {0.7912} {0.9392} {0.2003}** 0.7376*** 0.0034 0.0453 −0.0266} {0.0001} {0.9706} {0.6807} {0.8439}

(continued on next page)

Table 7Japanese information spillovers—dynamic

For Australia, HK and SP: RAPID,t = αR_JP· RJP

ID,t−1 + αVM_JP· VMJPID,t−1 + M(·), lnhAP

ID,t = βVT_JP

For the US: RUSID,t = αR_JP· RJP

ID,t + αVM_JP· VMJPID,t + M(·), lnhUS

ID,t = βVT_JP· VTJPID,t + βVM



αc −0.0015 −0.0025 −0.0056 −0.0507*** −0.0159 0.0042 0.0175*** −0.0077{0.3496} {0.1266} {0.6217} {0.0014} {0.2452} {0.8547} {0.0000} {0.0659

αHol −0.0006 0.0007 0.0014 0.0265*** 0.0081* 0.0257 −0.0083* −0.0006{0.5415} {0.3738} {0.9060} {0.0049} {0.0727} {0.1546} {0.0550} {0.9420

αR_JP 0.0007** −0.0055** −0.0007*** 0.0074*** 0.0021 0.0543*** −0.0166*** −0.0136{0.0430} {0.0017} {0.0000} {0.0061} {0.6300} {0.0044} {0.0000} {0.0002

αVM_JP −0.0070** 0.0138*** −0.0036 0.0183 0.0208 −0.0278 0.0705*** −0.0654{0.0402} {0.0000} {0.2737} {0.2625} {0.2408} {0.1454} {0.0000} {0.0000

βc −0.4576*** −0.0293 −3.6944*** −0.0191 0.1210 −0.2235*** −0.3119* 0.0126{0.0079} {0.8835} {0.0000} {0.8370} {0.2708} {0.0002} {0.0778} {0.8476

βε1 −0.0186 −0.0547 0.5979*** −0.0869*** −0.0536* −0.1693** −0.1095** −0.0630{0.7002} {0.2736} {0.0057} {0.0012} {0.0546} {0.0321} {0.0148} {0.0700

βε2 0.2984*** 0.4110*** 0.9442*** 0.1919*** 0.0781 0.6659*** 0.4618*** 0.0941{0.0000} {0.0001} {0.0000} {0.0003} {0.1619} {0.0000} {0.0000} {0.0035

βh 0.9301*** 0.8784*** 0.1994*** 0.9590*** 0.9696*** 0.0963 0.8708*** 0.9877{0.0000} {0.0000} {0.0078} {0.0000} {0.0000} {0.6043} {0.0000} {0.0000

βHol 0.2109** −0.2001 0.6784*** 0.0101 −0.1030 0.2472*** 0.1449 −0.0142{0.0138} {0.1265} {0.0000} {0.8594} {0.1251} {0.0000} {0.1127} {0.6915

βVT_JP 0.5169* 0.7336*** −4.2821*** 0.1009 −0.0085 −0.3481 0.4849 0.1822{0.0636} {0.0066} {0.0000} {0.5101} {0.9577} {0.2687} {0.1393} {0.1531

βVM_JP 0.6541* −0.4189 0.8886 0.8962*** 0.8469*** 0.6518** 0.5876*** 0.6157{0.0615} {0.1972} {0.4466} {0.0000} {0.0000} {0.0285} {0.0002} {0.0478

362S.-J.K

im/J.Japanese

Int.Econom

ies19

(2005)338–365

US


10 6 6 60.28 −0.37 −0.29 −0.21

12.27 1.86 1.69 0.8129.68* 20.52 15.44 20.25

} {0.0753} {0.4261} {0.7507} {0.4425}11.72 13.58 15.16 11.12

} {0.9255} {0.8512} {0.7669} {0.9431}0.37 7.29* 2.72 1.33

} {0.9460} {0.0631} {0.4363} {0.7210}

Table 7 (Continued)



q 6 6 4 4 4 10 10Sk-z −1.7744 −1.58 −0.19 −0.09 0.06 −0.15 −1.57 −3.45Kur-z 53.5894 30.2804 53.55 13.03 12.05 3.61 35.21 46.11Q(20)-z 54.99*** 23.47 73.53*** 11.54 12.17 22.49 30.01* 26.62

{0.0000} {0.2662} {0.0000} {0.9311} {0.9102} {0.3147} {0.0698} {0.1463Q(20)-z2 133.46*** 21.78 50.92*** 10.21 17.78 27.42 5.62 7.49

{0.0000} {0.3528} {0.0002} {0.9643} {0.6016} {0.1239} {0.9993} {0.9947E–N 0.5643 2.81 0.15 5.87 7.53* 5.26 3.90 7.21*

{0.9046} {0.4217} {0.9851} {0.1180} {0.0568} {0.1540} {0.2728} {0.0655



in alln flows

adingffectsnse ine USecond,oth

h con-nd notuence

milar

k mar-overscorrela-in all

le indi-y tests

addi-nger

infor-ificantpanesee vol-s hadly andorane-neous

lloverllovere con-sitiveerally

rkets.fter thence ofin the

Japanese markets. The volume to volatility spillovers, however, are generally positivemarkets and are larger in magnitude compared to the contemporaneous informatioreported inTable 5, as in the case of the dynamic US spillovers reported above.

The investigations of the contemporaneous and dynamic return, volatility and trvolume spillover effects reveal the followings. First, the contemporaneous spillover eof returns from both markets are positive and significant, and they are more intethe post-1997 period in all cases. The dynamic return to return spillovers from thare still positive and significant, but the Japanese information had mixed results. Sthe volatility spillover effects are generally positive for the US information, while bpositive and negative Japanese volatility influences are found. Third, although bottemporaneous and dynamic US volume effects on the returns are generally weak aconsistent throughout the markets, there is a clear evidence of significant positive inflof dynamic volume information on the conditional volatilities across the markets. Siresults are observed for the Japanese volume information spillovers.

6. Conclusion

This paper investigated the nature of linkages of the advanced Asia–Pacific stockets of Australia, Hong Kong, Japan, Singapore with the US, and information spillfrom the US and Japan to the other markets. The analyses of contemporaneoustions of daily market returns show significant first and second moment correlationscountry pairs, and these correlations are significantly higher in the post-1997 sampcating more intensified market linkages after the crisis period. The Granger causalitrevealed that it is the US market that Granger caused the Asia–Pacific markets. Intion to the return to return and volatility to volatility causations, US volume also Gracaused both the returns and volatilities of the Asia–Pacific markets. The Japanesemation, however, failed to Granger cause the other Asia–Pacific markets to any signextent, although the US market return and volatility were Granger caused by the Jareturn and volatility, respectively with essentially no causal flow from the Japanesume data. Finally, the investigations of the spillover effects reveal that the US returna significant influence on the returns of the other markets both contemporaneouswith a lag, while the Japanese returns had the similar positive spillovers contempously but lagged information had mixed results. In both cases, only the contemporaspillovers are considerably larger in the post-1997 sample. While the US volatility spiis generally positive and significant, especially with a lag, the Japanese volatility spieffects are inconsistent across markets. The volume information spillovers on thditional volatilities of the other markets, however, are generally significant and poespecially with a lag. However, the volume spillover effects on the returns are genweak and not uniform across the markets.

In short, a complex array of market linkages exists in the Asia–Pacific stock maAlthough the contemporaneous market linkages in the region tended to be higher a1997 Asian crisis period suggesting increasing market integration, there is no evidemore intensified dynamic information spillover from the US and Japanese markets
later sample. The Asia–Pacific markets responded uniformly to the US information flows,


ly weakarketsand

egion,ges

ntial

October

US and

tween

r. Dice

nce 49

pirical

a–Pacific

g equity

volatil-

55, 913–

ACAP

rsion of

1778.

7–78.mpirical

arket

tional

arkets.

in general, and the Japanese information spillovers are market specific and generalcompared to the US influence. This might be explained by the fact that the regional mreceive significant portfolio investment inflows from the US thus exhibiting significantconsistent information spillovers from the US. The Japanese financial flows to the rhowever, are mostly via bank lending which would result in indirect financial linkabetween the Japanese and the regional stock markets.

Acknowledgment

I thank Takeo Hoshi (Editor) for his valuable comments, which led to a substaimprovement in the paper. The remaining errors, if any, are my own.

References

Arshanapalli, B., Doukas, J., 1993. International stock market linkages: evidence from the pre- and post-1987 period. J. Banking Finance 17, 193–208.

Arshanapalli, B., Doukas, J., Lang, L., 1995. Pre- and post-October 1987 stock market linkages betweenAsian markets. Pacific-Basin Finance J. 3, 57–73.

Bae, K.H., Karolyi, A., 1994. Good news, bad news and international spillovers of stock return volatility beJapan and the US. Pacific-Basin Finance J. 2, 405–438.

Bae, K.H., Karolyi, A., Stulz, R., 2000. A new approach to measuring financial contagion. Working papeCenter at Fisher College of Business, Ohio State University.

Blume, L., Easley, D., O’Hara, M., 1994. Market statistics and technical analysis: the role of volume. J. Fina(1), 153–181.

Bollerslev, T., Chou, R.Y., Croner, K.F., 1992. ARCH modelling in finance: a review of the theory and emevidence. J. Econometrics 52, 5–59.

Cha, B., Cheung, Y.-L., 1998. The impact of the US and the Japanese equity markets on the emerging Asiequity markets. Asia–Pacific Finan. Markets 5, 191–209.

Cha, B., Oh, S., 2000. The relationship between developed equity markets and the Pacific Basin’s emerginmarkets. Int. Rev. Econ. Finance 9, 299–322.

Chen, G.M., Firth, M., Rui, O.M., 2001. The dynamic relation between stock returns, trading volume, andity. Finan. Rev. 36, 153–173.

Chordia, T., Swaminathan, B., 2000. Trading volume and cross-correlations in stock returns. J. Finance935.

Chow, H.-K., 1999. Co-movement in Asian stock indices and the impact of the Asian financial crisis. In: PConference, July 1999, Singapore.

Connolly, R.A., Wang, F.A., 2000. On stock market return co-movements: macroeconomic news, dispebeliefs, and contagion. Working paper. Rice University.

Engle, R.F., Ng, V.K., 1993. Measuring and testing the impact of news on volatility. J. Finance 48, 1749–Gallant, A.R., Rossie, P.E., Tauchen, G., 1992. Stock prices and volume. Rev. Finan. Stud. 5, 199–242.Garman, M., Klass, M., 1980. On the estimation of security price volatility from historical data. J. Bus. 53, 6Ghosh, A., Saidi, R., Johnson, K.H., 1999. Who moves the Asia–Pacific stock markets—US or Japan? E

evidence based on the theory of cointegration. Finan. Rev. 34, 159–170.Girard, E.C., Rahman, H., 2002. The effect of the Asian financial crisis on stock returns, volatility and m

integration in the region. Stud. Econ. Finance 20, 51–75.Hamao, Y.R., Masulis, R.W., Ng, V.K., 1990. Correlations in price changes and volatility across interna

stock markets. Rev. Finan. Stud. 3, 281–307.Janakiramanan, S., Lamba, A., 1998. An empirical examination of linkages between Pacific-Basin stock m

J. Int. Finan. Markets, Inst. Money 8, 155–173.


oney

nal. 22,

urn co-

oney

stic and

m the

rkets.

ce 19,

arkets.

kages.

. J. Int.

rica 51,

er em-

Kaminsky, G., Schmukler, S.L., 1999. What triggers market jitters? A chronicle of the Asian crisis. J. Int. MFinance 18, 537–560.

Karpoff, J.M., 1987. The relation between price changes and trading volume: a survey. J. Finan. Quant. A109–126.

Karolyi, G.A., Stulz, R., 1996. Why do markets move together? An investigation of US–Japan stock retmovements. J. Finance 51, 951–986.

Koutmos, G., Booth, G., 1995. Asymmetric volatility transmission in international stock markets. J. Int. MFinance 14, 747–762.

Lee, B.-S., Rui, O.M., 2002. The dynamic relationship between stock returns and trading volume: Domecross-country evidence. J. Banking Finance 26, 51–78.

Liu, Y.A., Pan, M.-S., Fung, H.-G., 1996. International transmission of stock price volatility: evidence froUS and six Pacific Basin markets. J. Multinat. Finan. Manage. 6, 81–94.

Liu, Y.A., Pan, M.-S., 1997. Mean and volatility spillover effects in the US and Pacific-Basin stock maMultinat. Finance J. 1, 47–62.

Ng, A., 2000. Volatility spillover effects from Japan and the US to the Pacific Basin. J. Int. Money Finan207–233.

Pan, M.-S., Liu, Y.A., Roth, H.J., 1999. Common stochastic trends and volatility in Asian–Pacific equity mGlobal Finance J. 10, 161–172.

Phylaktis, K., 1997. Capital market integration in the Pacific Basin region: an analysis of real interest rate linPacific-Basin Finance J. 5, 195–213.

Phylaktis, K., 1999. Capital market integration in the Pacific Basin region: an impulse response analysisMoney Finance 18, 267–287.

Tauchen, G.E., Pitt, M., 1983. The price variability–volume relationship on speculative markets. Economet485–505.

Theodossiou, P., Lee, U., 1993. Mean and volatility spillovers across major national stock markets: furth
pirical evidence. J. Finan. Res. 16, 337–350.